An experimental study of the effects of large distributed roughness located near the leading-edge of an airfoil has been performed to determine the effect on boundary-layer development and transition. Boundary-layer measurements were carried out on a two-dimensional NACA 0012 airfoil with a 21" chord. The distributed roughness used was in the form of densely packed large hemispherical elements applied to the leading-edge region of the airfoil section. The size of the distributed roughness was greater than or on the order of the laminar boundary-layer thickness. The roughness used is of the type and density observed to occur during the initial glaze ice accretion process. Detailed boundary-layer measurements were obtained through the use of hot-wire anemometry at Reynolds numbers of 0.75 $times$ 10$sp6$, 1.25 $times$ 10$sp6$, and 2.25 $times$ 10$sp6$. These measurements included mean and fluctuating velocity, turbulence intensity, flowfield intermittency, frequency content, and associated integral parameters. Both the clean model and roughness induced transitional boundary-layers were studied in great detail. Results from this investigation have shown that the transitional boundary-layer induced by large distributed roughness is markedly different from the clean model Tollmein-Schlicting induced transition process. For the type of roughness used in this study no fully developed turbulent boundary-layers were observed to occur at the roughness location, even though roughness heights were substantially greater than empirically determined critical roughness heights required to cause transition. Instead, the large distributed roughness was observed to trigger a transitional boundary-layer at or very near the roughness location. This transitional boundary-layer required a substantial chordwise extent to obtain a fully developed turbulent state. Turbulence intensity levels in the roughness induced transitional region were observed to be relatively low as compared to the clean model transitional region. In general, the clean model transitional process was substantially more energetic than that observed for the roughness induced boundary-layer. A new method for determining the location and extent of the transitional region by integrating the profiles of intermittency is also presented.